30 Seconds to a Live Landing Page: My First Night with the x-rush Text Module

The 30-second version

I'd heard of x-rush.ai but never opened it. Figured it was another meme-image thing. I was wrong.

I registered with email - no card, no phone, just in. The workbench has four module tabs along the top. I clicked Text. That's when the tab lights up teal - #00D4AA, the module's accent color - and you realize the whole text surface is keyed to that one shade. Sounds cosmetic, but it orients you fast: teal means "you're in text land."

Left nav: HTML Pages. I clicked it. A grid of templates. I filtered by Featured (hit ⌘K and the search box auto-focuses, nice touch), picked one called "Product Hero" - dark, centered headline, two buttons, a screenshot placeholder. Hit Preview. Looked clean. Hit Use Template.

Now I'm in the editor. Two panes: code on the left, live preview on the right. The preview has four device tabs - Desktop, Tablet (768), Mobile (375), and Code - so I flicked to Mobile to make sure the stack didn't collapse. I clicked the headline - "Your Product Name" - and typed "Lunchbox." The preview updated before my finger left the key. No rebuild, no spinner. That's the moment. You stop thinking "will this render" and just type.

I swapped the subhead, changed the two button labels (Get Lunch / See How), pasted a screenshot URL into the placeholder. Maybe 90 seconds total - I'm rounding down for the title, sue me. Then Ctrl+S. A toast: "Saved." Then Publish. Another toast, and a URL: x-rush.ai/p/lunchbox-7f3a.

One nice detail: x-rush keeps your last 10 publish snapshots. An hour later I bungled the button colors, hit rollback to the 9am version, done - no re-editing. I texted the URL to two friends.

One replied: "Nice, you spent all afternoon on this?" No. I spent it eating dinner.

When I actually needed help: the AI assistant

Saturday morning I wanted a contact form. I don't write forms often enough to remember the right input types, and I definitely didn't want to hand-style validation. This is where I expected to bounce off the tool.

Instead I opened the AI assistant panel - right side of the editor, a chat input. I typed: "add a contact form below the hero, name + email + message, validate email, submit shows a thank-you."

The assistant had already read the current file (it said so explicitly: "looking at your current HTML..."). It came back with a diff - green lines for added, red for removed, the way git shows it. Not a wall of code dumped on me. I could read exactly what changed, line by line.

I hit Apply. The form appeared in the preview. The email validation worked. I hadn't written a line.

Here's the part that mattered: I didn't like the field spacing on a follow-up. I hit Reject, then Ctrl+Z a couple times - and I was back to the clean original template. The undo chain goes all the way back, so the AI can't brick your starting point. That's the difference between "AI edits your file" and "AI edits your file safely." A lot of tools get the first half right and skip the second.

(For the curious: x-rush plugs into the industry's top text models - Claude Fable 5, GPT-5.6, Qwen3.7 Max, and friends - and smart-routes each task to the one that fits. The diff felt instant because routing picks a fast model for a page edit. The pool refreshes as the world moves; whatever ships next, the router uses it where it wins.)

Auto Content: the nudge that doesn't nag

I'm easily annoyed by notifications, so I almost skipped Auto Content. Glad I didn't.

I picked two agents: a daily tarot pull and a "weird food trivia" one (felt on-theme for Lunchbox). Set them to push once a day. The thing I like: inside the workbench (/workbench/*), the global chat bubble is hidden. So while I'm actually working, nothing pings me. The pushes land in a feed I check when I want to, not when they demand it.

And when a trivia push is genuinely interesting - "the first household refrigerator cost more than a car in 1922" - there's a "continue chatting" link that jumps me into AI Chat with that agent and the message already loaded. The push is a hook, not a dead end.

AI Chat: I built a code reviewer

This is where I got hooked for real.

AI Chat lets you define your own agent. I made one called "Grumpy Reviewer." Personality: "terse, skeptical, assumes I'm wrong until proven otherwise." Background prompt: the stack I work in, the fact that I optimize for readability over cleverness, and a standing instruction to flag anything that looks copied from Stack Overflow without understanding.

I pasted the Lunchbox page's JS into it. It came back with three complaints, all fair. (The sarcastic tone was my fault - I asked for it.)

The detail I didn't expect: each agent keeps its own history. I switch from Grumpy Reviewer to the Tarot Reader and back, and the Grumpy conversation is exactly where I left it. Per-agent memory means you don't lose context when you hop around. Most chat tools reset on switch. This one doesn't.

The loop I ended up in

By Sunday I wasn't building a landing page anymore. I'd fallen into a loop:

• T-2 (Auto Content): a daily push drops a weird food fact into my feed.
• T-3 (AI Chat): I take the interesting ones into a deep chat with a custom agent, and a post idea takes shape.
• T-1 (HTML Pages): I lay it out in the editor, hit Publish, get a URL.

That's the real pitch - not "make a page fast" but "a daily current that keeps producing things worth laying out." The 30-second landing page is the door. The loop is the room.

Where to go deeper

If any of this landed, the docs go further than I have room for:

• Text module overview - the full surface, not just the 30-second tour
• Smart page generator - how the templates and AI assistant actually fit together
• Auto content generator - the push system and the workbench bubble rules
• AI characters & chat - building and managing custom agents

I came in needing a landing page for a hackathon I half-finished. I stayed because the loop is genuinely useful on a random Tuesday. Your mileage may vary - but the 30 seconds are free, and that's a cheap way to find out.